Lecture 6 Classical Control Overview IV Dr. Radhakant Padhi Asst. Professor Dept. of Aerospace Engineering Indian Institute of Science - Bangalore
Lead Lag Compensator Design Dr. Radhakant Padhi Asst. Professor Dept. of Aerospace Engineering Indian Institute of Science - Bangalore
Motivation The PID controller involves three components: Proportional feedback Integral feedback Derivative feedback Problem in PID design: Requirement of pure integrators and pure differentiators, which are difficult to realize Pure integrator pole may travel to the right half plane because of realization inaccuracies. E( s) U( s) Question: Can the difficulties of the PID design be avoided, without compromising much on the basic design philosophy? Answer: YES! Through Lead- Lag compensator design. 3
Ideal Integral Compensation for Improving SS Error: PI Controller Point A: Desired pole location (already satisfactory transient response). No need to change its location. However, need to improve the steady-state error. K ( s + a) Ka K I = K + = K + (PI) s s s 4
Lag Compensator (for Improving Steady State Error) K v 0 = Kz1z2 p p 1 2 Note: Pole-zero pair should be close to each other 1 0 ( ) K v 1 = ( Kz1z2 )( zc ) ( p p )( p ) 1 2 z c = Kv K 0 > v0 p c (provided z > p ) K K, provided z / p 1. Hence put the pair close to origin. v v c c c like PI controller c c 5
Lag Compensator (for Improving Steady State Error) Root locus before compensation Root locus after compensation No appreciable change in Root locus & closed loop pole location: No difference in transient response, but improvement in SS error! 6
Lead Compensator (for Improving Transient Response) Objective: To improve transient performance, avoiding the pure PD realization Advantages; Avoids realization difficulties (e.g. avoids requirement of additional power supplies in electrical circuits) Reduces noise amplification due to differentiation Drawback: Addition of a pure zero in PD controller tends to reduce the number of branches of Root Locus that travel to RH plane, whereas Led compensators are not capable of doing that. 7
Geometry of Lead Compensator (Root locus of original system does not pass through it) θ θ θ θ + θ = (2k + 1)180 2 1 3 4 5 where ( θ θ ) = θ : Angular contribution of compensator 2 1 c 0 8
Many Possibilities of Lead Compensators Different selections results in: Different gain values to reach the desired point Different static error constants (that leads to different SS errors); hence different closed loop response in strict sense! 9
Lead-Lag Compensator Design Design Steps: First, evaluate the performance of the uncompensated system. If necessary, design a lead compensator to improve the transient response. Next, design a lag compensator to improve the steady state error. Simulate the system to be sure that all requirements have been met. Redesign the compensators (i.e. retune the compensator gains), if the simulation performance in not satisfactory. Ref: N. S. Nise: Control Systems Engineering, 4 th Ed., Wiley, 2004 10
Frequency Response Analysis Dr. Radhakant Padhi Asst. Professor Dept. of Aerospace Engineering Indian Institute of Science - Bangalore
Concept of Frequency Response ( )( ) M ( ω) φ ( ω) = M ( ω) φ ( ω) M( ω) φ( ω) 0 0 i i SS output sinusoid Frequency Response Input sinusoid [ ] = M ( ω) M( ω) φ ( ω) + φ( ω) M ( ω) = i M M 0 0 ( ω) ( ω) φ( ω) = φ ( ω) φ ( ω) i i i 12
Concept of Frequency Response rt = A t+ B t= A + B t B A 2 2 1 ( ) cosω sinω cos ω tan ( / ) ( ) = M t M = A + B = B A 2 2 1 icos ω φi, where i, φi tan ( / ) After appropriate analysis: c () t = M M cos t+ + 0 0 ( ω φ φ ) ( )( ) ss i G i G M φ = M φ M φ i i G G where G ( ω), φ ( ω) M = G j = G j G 13
What is frequency response? Magnitude and phase relationship between sinusoidal input and the steady state output of a linear system is termed as frequency response. Commonly used frequency response analysis: Bode plot Nyquist plot Nichols chart Cs () T() s =, s = jω, Rs () C( jω) T( jω) = = M φ R( jω) M = T( jω), φ = T( jω) 14
Bode Plot Analysis Dr. Radhakant Padhi Asst. Professor Dept. of Aerospace Engineering Indian Institute of Science - Bangalore
Bode Plot Analysis Bode plot consists of two simultaneous graphs: Magnitude in db (20 log G(jω) ) vs. frequency (in logω) Phase (in degrees) vs. frequency (in logω) Steps: K( s+ z1)( s+ z2) ( s+ zk ) Gs () = m s ( s+ p )( s+ p ) ( s+ p ) 1 2 Then 20log K + 20log ( s+ z1) + + 20log ( s+ zk ) 20log G( jω) = m 20log s 20log ( s+ p1 ) 20log ( s+ p ) m ( 1 k ) ( 1 k ) G( jω) = ( s+ z ) + + ( s+ z ) s + ( s+ p ) + + ( s+ p ) n n s jω s jω 16
Bode Diagrams: Advantages All algebra is through addition and subtraction, and that too mostly through straight line asymptotic approximations Low frequency response contains sufficient information about the physical characteristics of most of the practical systems. Experimental determination of a transfer function is possible through Bode plot analysis. 17
Bode Diagrams In Bode diagrams, frequency ratios are expressed in terms of: Octave: it is a frequency band from ω 1 to 2ω 1. Decade: it is a frequency band from ω 1 to 10ω 1, where ω 1 is any frequency value. The basic factors which occur frequently in an arbitrary transfer function are: Gain K Integral and derivatives: ( jω ) ± 1 1 + jω T ±, T = First order factors: Quadratic Factors: ( ) 1 1 a + ξ( j ω ω ) + ( jω ω ) ( 2 ) ± 1 2 1 n n 18
Bode Diagrams For Constant Gain K, log-magnitude curve is a horizontal straight line at the magnitude of (20 log K) db and phase angle is 0 deg. Varying the gain K, raises or lowers the logmagnitude curve of the transfer function by the corresponding constant amount, but has no effect on the phase curve Logarithmic representation of the frequencyresponse curve of factor ( j( ω / a) + 1) can be approximated by two straight-line asymptotes Frequency at which the two asymptotes meet is called the corner frequency or break frequency. 19
Example 1: Gs () = s+ a, G( jω) = jω + a At low frequencies, Magnitude/Phase response: At high frequencies, Magnitude/Phase response: ω << a, G( jω) a 20log M = 20log a, G( jω) = 0 ω >> a, G( jω) jω 0 0 20log M = 20log ω, G( jω) = 90 ω G( jω) = jω+ a = a j + 1 ω = a a Corner frequency: c 20
Bode Plot for (s +a) 21
Bode diagrams of some standard first order terms Ref: N. S. Nise, Control Systems Engineering, 4 th Ed. Wiley, 2004. Bode plots for: a. G(s) = s b. G(s) = 1/s c. G(s) = (s + a) d. G(s) = 1/(s + a) 22
Example 2: Gs () = Ks ( + 3) ss ( + 1)( s+ 2) 23
Example 2: Gs () = Ks ( + 3) ss ( + 1)( s+ 2) 24
First Order Terms: Comparison of Actual and Asymptotic Behavior Magnitude comparison Phase comparison Ref: N. S. Nise, Control Systems Engineering, 4 th Ed. Wiley, 2004. 25
Bode plot for second order systems System with conjugate zeros when 2 2 Gs () = s + 2ξωns+ ωn System with conjugate poles when Gs () = s 1 2 2 + 2ξωns+ ωn For complex conjugate poles and zeros, the slope changes by 40dB/decade ± 0< ξ< 1 0< ξ< 1 26
For low frequencies (ω<<ω n ), Log magnitude becomes 0dB (20log 1= 0), hence low frequency asymptote is a straight horizontal line For high frequencies (ω>>ω n ), the log magnitude becomes High frequency asymptote is a straight line with slope of -40dB/decade The phase angle of the quadratic factor is 27
Scaled Response for Magnitude Plot Phase Plot 28
Scaled Response for Magnitude Plot Phase Plot 29
Example 3: Gs () = ( s + 3) + + + 2 ( s 2)( s 2s 25) Second order system is normalized Gs () = s 1 3 + 3 + 1 + s + 1 2 25 25 2 50 s s 2 Bode magnitude plot starts from 20logK = 24.44dB and continues until the next corner frequency at 2 rad/s 30
Magnitude Plot 31
Phase plot 32
Nyquist Plot Analysis Dr. Radhakant Padhi Asst. Professor Dept. of Aerospace Engineering Indian Institute of Science - Bangalore
Presenting frequency response characteristics (Polar plots) Frequency response of transfer functions can be represented by Polar plots, also called Nyquist plots, as shown in the figure. M = G( jω) φ = G( jω) 34
General nature of Nyquist curves Ref: K. Ogata: Modern Control Engineering, 3rd Ed., Prentice Hall, 1999. 35
Nyquist plots for several standard transfer functions Ref: K. Ogata: Modern Control Engineering, 3rd Ed., Prentice Hall, 1999. 36
Nyquist stability criterion Special case: (G(s)H(s) has neither poles nor zeros on the jω axis) If a contour, that encircles the entire right half plane is mapped through G(s)H(s). Then number of closed loop poles, Z, in right half of s-plane equals the number of open loop poles P, that are in right half of s-plane minus the number of counter-clockwise revolutions N around -1 of the mapping, i.e. Z = P - N. a. contour does not enclose closed-loop poles b. contour does enclose closed loop poles 37
Example 1 Open loop poles in the right half of s- plane are 2,4, i.e, P = 2 Number of encirclements of (-1), N = 2 Z = P-N = 0, hence the system is stable. 38
Example 2: Gs () K ss ( + 3)( s+ 5) There are no open loop poles in the right half of s-plane, i.e, P = 0. Number of encirclements N = 0. Z = P - N = 0, hence the system is stable. Value of K which determines the stability is 120.5. It implies if K < 120.5 then system is stable. if K > 120.5, critical point is encircled and N = -1. In that case Z = P - N = 1, and hence the system is unstable = 39
Robustness Concepts Dr. Radhakant Padhi Asst. Professor Dept. of Aerospace Engineering Indian Institute of Science - Bangalore
Definitions The closed loop poles can be determined through characteristic equation 1 + GsHs ( ) ( ) = 0, s= jω GsHs () () = 1+ j0= 1 180 Phase cross over frequency (ω pc ) 0 Frequency at which the phase angle of the transfer function becomes -180 o 0 G( jω) H( jω) = 180 Gain cross over frequency (ω gc ) Frequency at which the magnitude of the open loop transfer function, is unity, i.e. IG(jω)H(jω) =1. These frequencies play an important role in determining the stability margins of the system. 41
Gain and Phase Margins through Nyquist Plots γ 0 = 180 φ K g = 1 G( jω) System is said to be Stable, if G M and Φ both are positive, i.e. M ω pc > ω gc Marginally stable, if G M and Φ both are zero i.e. ω M pc = ω gc Unstable, if G M and Φ both are negative i.e. M ω pc < ω gc 42
Gain and Phase Margins through Bode Plots For a stable minimum-phase system, GM/PM indicates how much gain/phase can be increased before the system becomes unstable. For an unstable system, GM/PM is indicative of how much gain/phase must be decreased to make the system stable. 43
Frequency Response Characteristics Consider a closed loop transfer function of second order system Gs () = s 2 n 2 2 + 2ξωns+ ωn The closed loop frequency response ω The frequency at which the M reaches its peak value is called resonant frequency (ω P ) 2 ω = ω 1 2ξ At ω P, the slope of the magnitude curve is zero. P n 44
Frequency Response Characteristics The maximum value of magnitude is known as the resonant peak (M p ) M P = 1 2 2ξ 1 ξ Bandwidth (ω BW ) is the frequency at which the magnitude response curve is 3dB down from its value at zero frequency. ( ) 2 4 2 ωbw = ωn 1 2ξ + 4ξ 4ξ + 2 f ( ξ ) T 4/ ξ / ω s ( ) n 45
References (Classical Control Systems) N. S. Nise: Control Systems Engineering, 4 th Ed., Wiley, 2004. K. Ogata: Modern Control Engineering, 3 rd Ed., Prentice Hall, 1999. J. J. Distefano III et al.: Feedback and Control Systems, 2 nd Ed., Mc Graw Hill, 1990 (Schaum s Outline Series). B. N. Pamadi: Performance, Stability, Dynamics and Control of Airplanes, AIAA Education Series, 1998. R. C. Nelson: Flight Stability and Automatic Control, AIAA Education Series, 1998. E. Kreyszig: Advanced Engineering Mathematics, 8th Ed., Wiley, 2004. 46
Some Points to Remember Dr. Radhakant Padhi Asst. Professor Dept. of Aerospace Engineering Indian Institute of Science - Bangalore
Some Points to Remember Block diagrams are not circuit diagrams! Stability and robustness are necessary of any good control design. After that, one must look for performance (fast response, optimality etc.) High controller gains: Good benefits Robust stability, Good tracking Bad effects Control saturation, Noise amplification Inter-coupling of variables, nonlinearities, control and state saturation limits, time delays, quantization errors etc. are always present: Classical SISO approach may not be adequate; Advanced techniques are needed. 48
49
50